Technical Proposal Title: Characterization and Physics-based Modeling of Electrochemical Memristors Principal Investigator: Hugh Barnaby, Associate Professor, Electrical Engineering, Arizona State University Abstract- This two year program will focus on developing and validating models that describe the underlying physics governing the electrical behavior of electrochemical memristors. Analytical models capturing charge transport and ion reaction properties in solid state electrolyte materials used for these devices will be developed and implemented in finite element solvers. The numerical simulations will enable a physics-based characterization of the dynamics of filament formation and dissolution in ion conducting films. Model parameters and electrical characteristics will be obtained from and validated with experiments on test structures designed and fabricated during the program. The project will lead to significant advances in our understanding of the physics of electrochemical memristor operation, support the identification of potential radiation threats as well as strategies for improving the reliability and hardness of these technologies. Introduction- Silver-based electrochemical memories show enormous potential for applications in non-volatile memory, as well as in more challenging, mixed signal applications, including neuromorphic computing. These devices are fabricated with thin films of chalcogenide glass (ChG) ion conductors sandwiched between active and inert metallic electrodes. Through a combination of reduction/oxidation (ReOx) and Ag+ ion transport processes, conducting filaments in the ChG film may be grown or dissolved as a function of bias, thereby enabling resistive switching between the two terminals. While several groups have made significant strides in device development, and in process integration, to the point where this technology is approaching readiness in non-volatile memory applications, significant challenges remain to improve cell function, reliability, and cycling endurance, as well as to facilitate their practical use in analog applications. The central problem is the large variability, even within the same device, of operational parameters and programmed resistance. To understand these variabilities, we need to understand and model: 1) the solid state characteristics of the ChG film and the metal-glass interface; 2) charge transport and reaction properties in the ChG bulk and interfaces; and 3) the physics of conducting filament formation and dissolution. While continued engineering approaches could lead to technological breakthroughs, we contend that this problem would be better solved with a balanced approach that includes model-based theory with experimental studies on both isolated materials, and device structures. We already know that memristive action in these electrochemical devices is a function of ChG stoichiometry and of transition metals doped into the glass. However, at present there are no physical models that capture the process of resistive switching in a chemically specific way, and the experiments are controversial. We anticipate that the filament growth depends on the composition of the bottom electrode and properties of the glass film, but there is, to date, neither quantitative measurements nor detailed finite element simulations that capture this. In this program we plan to work closely with theorists on the molecular dynamics (MD) and first principles quantum mechanical defect studies of the active device materials to build models that incorporate detailed physics and to simulate how these devices work electrically, how they age, and how they respond in radiation and extreme temperature environments. Objectives- Recently, Monte Carlo simulation techniques have been developed to capture the kinetics of Ag transport and metallic filament formation in resistive memory engineered with chalcogenide films. In this program, the mechanisms of Ag transport and reactions will be modeled using device simulation. Specifically, the processes of oxidation, auto-ionization, field-induced ion transport, electro-deposition, and ion trapping will be modeled with a flux-based, drift-diffusion numerical solver that simulates the transport and reactions of ionic species in ChG solid state materials. Key questions that will be addressed in the program are: 1. What is the nature of bulk glass and the metal-glass interface? Specifically, can we quantify solid state parameters such as permittivity, affinity, density of states, and band-gap and can we determine how ion concentration (e.g., Ag) impacts these parameters. Moreover, what is the nature of the Ag-ChG interface and the glass interface with inert metal? Can it be thought of as a Schottky or tunnel contact? What roles do the metal work-functions and ChG material parameters play in ionic and electron/hole transport? 2. How is charge transport modeled and to what extent can standard flux-based functions such as drift-diffusion, Poisson, and continuity be used to describe memristor operation? What are the mechanisms Ag+ ion, neutral Ag, and charged carrier (electrons and holes) transport in ChG? Can ionic transport be modeled by a single mobility, or is a dispersive model required? Can we quantify parameters such as diffusivity and mobility of all charged species? Lastly, can the interaction of mobile ionic species with the host material be captured in continuity equations, relating recombination and generation to energy dependent reaction rates and flux gradients to ion and electron/hole transport properties? 3. What is the physics of metallic filament formation and dissolution? What is the static nature of Ag, in low and high impedance states? Is Ag growth in the ChG structure dendritic? Should fractal growth patterns be considered? How do growth and dissolution processes map to device operation and reliability? Lastly, how does Ag concentration impact the equivalent impendence of the system?
|Effective start/end date||9/18/13 → 9/17/15|
- DOD-USAF-AFRL: Air Force Office of Scientific Research (AFOSR): $148,878.00